At the OFC 2019 Media/Analyst Luncheon – “Hot Topics in Data Center Optics: Emerging Standards/MSAs, Silicon Photonics, 400G and Beyond,” Bill Gartner, Vice President/GM Optical Systems and Optics Group, Cisco Systems, mentioned the idea of “placing bets” because his company cannot get involved with everything in the space. In response to this point, this writer brought up that many people in the industry were scratching their heads about the Luxtera purchase, pointed out that there had been practically no public comments on silicon photonics technology by him over the last few years (other than apparently critical remarks), and sought out whether it would be used as a loss leader in the same way that Intel appears to be employing its sales of optical components. Gartner replied that the acquisition was not driven by the PSM4 product line, that Luxtera had an interesting roadmap as well as delivered on its past promises, and that it has made strides when it has come to testing, manufacturing, and assembly. In particular, he claimed that Cisco needed to have the technology because of Luxtera’s technical leadership position around embedded optics. Gartner further stated that his firm continues to invest in Lightwire, that the solution is used in several of its platforms, and that despite the previous market shift to QSFP28s, the supplier has not stopped shipping CPAKs. He also said that Cisco’s top management is emphasizing optics, along with software and silicon, as important areas of focus for the company’s future. Finally, Gartner seemed to go out of his way to assert that Cisco is the “biggest optics vendor on earth.” Regarding this last contention, we would not dispute it, given what we understand to be a considerable amount of OEMing of components by the vendor. Concerning the other justifications, fibeReality does not find them terribly convincing, particularly that optics will be stressed indefinitely as a driving force at the supplier.
For the present, Cisco has at least indirectly acknowledged that it will continue to ship PSM4 devices, which have, of course, become heavily challenged on their ability to make any money. In addition, while Luxtera is considered to be a comparatively good, SiPh manufacturer, and offers a relatively mature level of packaging in general, there are still the inherent hurdles with the technology as a whole, which the acquirer certainly knows intimately, and that it involves “fit[ting] a circle into a square.” Therefore, the theory of a “quid pro quo” with Microsoft ensuring the viability of Luxtera, while Cisco would ostensibly win back market share from Arista Networks at the operator’s data center account, continues to be logical to us.
Then there is the question -- why would Cisco otherwise really want to engage in the strange process of gluing a fiber array over a surface grating using a proprietary line, which cannot be transferred anywhere else? How would it not ultimately kill its cost model with any meaningful number of fibers, let alone result in a sufficiently attractive ROI, which evidently did not occur in picking up Lightwire.
We have discussed in the past that SiPh is not essential anyway, and is likely detrimental, if co-packaged optics actually comes to fruition. (Interestingly, and unsurprisingly, we found that Microsoft executives certainly strived as much as possible to avoid using the acronym, COBO, in their presentations at OFC.)
As always, fibeReality does not recommend any securities, and this writer does not invest in any companies being analyzed by us.
To follow us on our totally separate, quick update company blog, which is exclusively on fibeReality’s LinkedIn page, please click .
[written by Mark Lutkowitz]